This paper was originally presented at an Association of Research Libraries workshop, "The Specialized Scholarly Monograph in Crisis, or How Can I Get Tenure If You Won't Publish My Book?" at a session September 11, 1997, called "Economics of the Specialized Monograph." Copyright by Scott Bennett and reprinted here with permission. Readers of this essay may copy it without the copyright owner's permission if the author and publisher are acknowledged and the copy is used for educational, not-for-profit purposes.

This essay describes costs the library incurs in managing scholarly monographs that do not have many readers. My hope is to identify whether the much-vaunted advantages of just-in-time delivery might contribute to rescuing scholarly monographs from the difficulties of our traditional just-in-case methods of getting them into readers' hands.[1]

The typical just-in-case system for managing inventories involves a storage facility — be it a parts warehouse or a library stacks — serving both the manufacturer and the consumer, who in our case is a reader. By contrast, the just-in-time system connects the producer directly to the consumer, eliminating the costs of storage. The productive capacity to make copies of something, be it automobile dashboards or books, is linked directly to the consumer in a made-on-order transaction. So as a first approximation one might say the library costs for just-in-time electronic delivery of infrequently read monographs to readers should be zero, because the traditional library has no involvement in it. In describing this arrangement and the consequent shift in business away from libraries, Sandy Thatcher has written that "of course, this electronic alternative would require that some university resource be shifted away from libraries . . . and toward presses."[2] I will comment on this shift later, when I assess the political economy of just-in-time delivery, noting the shift would likely be to readers rather than to university presses.

Assuming in a second approximation that libraries might have an operational role in delivering infrequently read scholarly monographs, I looked for businesses actually doing that now. I focused on Ph.D. dissertations — one kind of little-read scholarly monograph — and on the Center for Research Libraries (CLR) as an agency that maintains a collection of foreign dissertations for just-in-case delivery to readers. By contrast, the business of UMI, Inc. turns on a productive capacity for the just-in-time delivery of copies of domestic dissertations. Happily, both are long-established successful businesses. Thanks to the ready cooperation of Donald Simpson, CRL president, I have some information about the Center's operating costs. For UMI, I rely on their price list.[3] I have supplemented information from these organizations with cost estimates for the Yale University Library and with the work Donald Waters has done in Yale's Project Open Book, modeling the cost structures of just-in-case and just-in-time delivery.

The argument that follows draws on cost models that are merely analytical constructs, using cost estimates more or less reliable. Moreover, the application of just-in-time delivery concepts to publishing and to library operations is itself highly problematic. And the technical construction of the models involves debatable propositions — as is always the case in modeling. All this being so, I emphasize that I do not offer these models as describing with precision a settled reality. My purpose is rather to use these models to speculate about some fast-evolving possibilities that information technology seems to offer. So please understand what follows to be an essay on what-ifs.

Modeled Costs

Taking UMI as an exemplar of just-in-time delivery of infrequently used monographs, I began with UMI's price for a softcover thesis, added their price for 5-6 day delivery (comparable to the delivery time from CRL), and subtracted the estimated cost of manufacturing the book, to arrive at my estimate of UMI's price of $47.50 per use for creating and maintaining a production capability to deliver copies of dissertations just when, and only when, someone wants them.[4] The document-delivery cost of $25 constitutes 53 percent of this per use cost. UMI's price does not vary with the frequency of use, and any number of uses can be supported essentially simultaneously.[5]

"The Project Open Book cost model might prompt one to embrace a digital, just-in-time model of operation"

Modeling the costs of traditional just-in-case library access to information is more complex. The traditional library does not make copies of books for each reader, but rather supports the shared use of a given book over time. Accordingly, traditional library access costs vary with frequency of use over time. In conceiving of the Center for Research Libraries as a just-in-case provider, I have therefore looked at two variables: the time period over which one depreciates the one-time costs of building the collection, and the frequency with which the book is used over that depreciation period. For the first of these variables, I included only the one-time costs of ordering and cataloging the material, and I have used 50-, 75-, and 100-year depreciation periods.[6] For the second set of variables, I included estimates for the sunk annual costs of shelving[7], preservation, and associated operations, and cost estimates for circulation services. Circulation costs include expedited long-distance delivery comparable to that provided by UMI's rush service.

Using those parameters, one would expect total unit costs to be driven down as depreciation periods lengthen and as use becomes more frequent. CRL's costs for building a comprehensive collection of foreign dissertations, preserving that collection over decades, and providing just-in-case access to it reflects those features of the model, with total per use costs estimated, in rounded numbers, between $34 and $21. Document-delivery costs of $20 constitute between 59 percent and 96 percent of this per use cost.

To identify more clearly the sensitivity of this model to individual cost factors, I used it with two somewhat different sets of figures that reflect the estimated operating costs of Yale University Library: those for conventional on-campus shelving complying with building codes accomodating readers with disabilities and those for high-efficiency off-campus shelving. The Yale figures include some direct costs that are included as indirect costs in the CRL figures. Using the same modeling parameters as were used for CRL, Yale's per-use costs for building a collection of books, preserving that collection over decades, and providing just-in-case access to it ranges, in rounded numbers, between $44 and $5 for conventional shelving, and between $27 and $4 per use for high-efficiency shelving. The document delivery cost of $1.75 constitutes between 4 percent and 45 percent of this per use cost.

Yale numbers are much lower than those for CRL and UMI principally because our unit costs for circulation are radically lower. Yale's paging costs are lower than CRL's because of different staffing patterns, and Yale incurs no expedited long-distance delivery expenses as both CRL and UMI do. Timeliness of access at Yale is for the most part secured by the reader's own labor, not by a commercial delivery service.[8] Otherwise,

  • the costs of ordering and cataloging are somewhat different between CRL and Yale, primarily because for Yale I have assumed a mix of original and copy cataloging. In fact, however, when these one-time costs are depreciated over 50 or more years, differences between them contribute very little to different outcomes.

  • there are several differences in the way CRL and Yale calculate and amortize space costs. These differences yield a lower cost/net square foot at CRL ($20.69, compared to Yale's $25.74), but the difference is lost in the much higher unit costs for circulation at CRL.

  • differences between conventional, on-campus ADA compliant bookshelves and high efficiency off-campus shelving drive significantly different capital and operating costs at Yale: $2.88 year/circulated volume for the former, compared to $1.27 year/circulated volume for the latter.

The UMI, CRL, and Yale activities modeled here employ nondigital means of providing access to infrequently read monographs. Donald Waters's pioneering work in Yale's Project Open Book models a digital alternative for storing and providing access to information.[9] Waters's model is, however, conceptually quite different from the one I have developed. I have focused on the costs of managing individual, infrequently read scholarly monographs. Waters modeled the costs of managing whole collections of imaged texts, including both high-use and low-use volumes. He also confined his cost model to two key variables, storage and access. And most important, he assumed that a digital library would store digitally only those titles that were actually used.[10] He compared the first-use storage and access costs of digital volumes to the cost of storing and providing access to whole collections of analog material, only some parts of which would actually be used.

Tables 1 and 2 detail the cost elements of a digital archive and a depository library, as modeled by Waters.[11] Table 3 shows what happens to those first-use costs over a ten-year period, assuming 13 percent annual decreases in digital hardware and software costs and 4 percent annual inflationary increases in other costs. In year one, the first use of a volume in the digital archive totals $9.30, or just more than twice the first use cost of $4.61 for a depository-library volume. But in this model all the costs of the depository library increase over time, with access (i.e., circulation) costs comprising about 95 percent of total first-use costs. By contrast, all the costs of the digital archive decrease over time, with access costs comprising between 65 percent and 77 percent of the total first-use cost. By the tenth year, initial positions are dramatically reversed. First-use costs in the depository library are $6.57, now nearly two times higher than the $3.61 first use cost of a digital-archive volume. The switch-over for those unit costs is in the fourth year. As Table 4 shows, when those unit costs are applied to entire collections, and when the key difference of the digital library storing only the volumes actually used is factored in, the digital archive becomes, in the seventh year, less expensive than the depository library. The cost difference favoring the first use of material in the digital archive is significant by the tenth year.

The Project Open Book cost model might prompt one to embrace a digital, just-in-time model of operation. That embrace must be conditioned by three problematic features of the model:

  • the conservative assumption that no productivity gains will be achieved over time in the operation of the depository library[12]

  • the assumption that all of the productivity gains provided by the digital archive's use of hardware and software will fall to the bottom line, with none of those gains being used by evolving systems designs that consume more computing resources or deliver new, higher quality functions

  • the fact that technology for the digital archive is subject to such fundamental change that its costs cannot be modeled for more than 10 years (if indeed for that long!), unlike those for the depository library where stable technologies allow one to model costs with more confidence for much longer periods of time.[13]

Those cautions are underscored when one focuses on the fact that in the Project Open Book model the $0.30 first use cost of holding a volume in a depository library remains much less expensive, even after ten years, than the $0.82 first-use cost of holding it digitally. It is the rapid fall in the cost of digital delivery, compared to the steady rise in the cost of delivering analog material, that so strongly favors the digital depository.

"There are hundreds of exemplars of just-in-time delivery of digital information"

As our topic is the infrequently read scholarly monograph, I have taken the liberty of moving Project Open Book's first use cost estimates (partially reported in Table 3) into my model. This is a considerable liberty given the quite different time horizons of the two models and the different assumptions made about what is shelved or stored. I have dealt with these difficulties by using unit costs (taken from Attachments 4 and 6) to model only the storage and access costs for a single book that is stored ten years and circulates only once at some point in those ten years. In all cases except Project Open Book's Digital Archive, the total of storage and access costs rises each year. In the case of the Digital Archive, the total falls each year. Even in Year 10, however, the total of digital access and storage costs is substantially higher than those for Project Open Book's Depository Library and higher than for the Yale Library estimates. Given the initial Year 1 spread between these costs and the modest slope of the trend lines, the two do not cross until several years beyond Year 10.[14]

Interpreting the Models

Overall, what do these cost models suggest? With regard to analog information resources, the cost difference between UMI and the Center for Research Libraries favors the Center's just-in-case model of operation. Even if one reduces UMI's price for just-in-time delivery by one-third, to address differences between for-profit and not-for-profit operations, the resulting $32 per use only brings the two operations into the same initial cost range. Just-in-case operations for analog material become decidedly attractive, as the Yale Library estimates suggest, if the collection can be shelved in a highly efficient manner relatively near to readers, so that both shelving and delivery costs can be held down.[15]

The digital arena for information is more complex. There are in fact hundreds of exemplars of just-in-time delivery of digital information. The networked databases and full-text resources that libraries never own but to which they now routinely license access are all just-in-time exemplars, as are the countless information resources freely available just-in-time via the World Wide Web. The principal direct cost to the library in providing such material is the license,[16] which compensates the publisher for a bundle of data acquisition and preparation, maintenance, and delivery costs. I can think of no actual just-in-time digital operation where this cost unbundling is done,[17] and so have depended instead on the Project Open Book model. That model identifies cost trends strongly favoring digital storage of and access to information, with crossover points for total storage and access costs happening sometime after the first 10 years. The Project Open Book model is fundamentally an argument that, with regard to the key factors of storage and access, it eventually will be cost effective to stop building local collections altogether and instead buy the information one needs only when one needs it. The economic rationale for the shared use of information — the economic rationale, that is, for traditional libraries — could disappear because the cost of managing and distributing that information to readers in the digital realm will be so low. With this change, traditional libraries might disappear, giving way to something like an ubiquitous, virtual bookstore for gaining access to information.

So finally might be realized the ambition of that remarkable early nineteenth century creation, the Society of Useful Knowledge, which pledged "to leave nothing undone, until knowledge has become as plentiful and as universally diffused as the air we breathe." So finally might be realized the pledge of the Society's publisher, Charles Knight, to work "till the wide fields of knowledge should become the inheritance of all."[18]

The Project Open Book model suggests we are some few years from seeing this happen, and I believe a realistic appraisal of several conditions built into the model will push the realization of the millennium of inexpensive storage of and access to information out a good many more years.[19] Even so, it may be useful to speculate briefly about the challenging political economy of the digital future that Project Open Book allows one to imagine, if not quite fully to believe in.

The Political Economy of Information Access

So long as the millennium is in front of us and access to digital information continues to carry a significant, rather than a trivial cost, [20] some means of rationing access will be needed. This is a problem of political economy. University-based readers of infrequently read scholarly monographs might, for instance:

  • purchase the information they need with personal funds, with no financial assistance from the university, or

  • purchase information with financial assistance, distributed to readers by a central administrative office or by the library's successor organization using some allocation rule (comparable perhaps to the rules used for travel funds).

The question before us is the market viability of the infrequently read scholarly monograph. We know from present experience that the first of these political economies — an unsubsidized free market for information — is unlikely to help the infrequently read scholarly monograph unless the cost of such books can be driven very far down. The second of these political economies requires a subsidy reflecting some policy view of the value of infrequently read books.[21] It also requires a set of allocation rules capable of managing several tangled questions. For instance, are full professors and professors active in research to be treated differently from or the same as assistant professors and professors whose research is quiescent? How does one balance research and teaching needs? Do different disciplines — physics and classics, for instance — have different cost bases? How might one calculate them? How are students supported? Are undergraduate and graduate students supported differently? Does the squeaky wheel get more resources? Who has the authority to make allocation decisions stick?

We have managed these problems with some success through the shared-information resources known as traditional libraries. We have carried these allocation solutions forward into the digital arena in the licenses libraries negotiate, on behalf of the communities they serve, for networked information. If, however, we make the next logical move in the digital environment, where digital copies of material are created, delivered, and paid for just when, and only when, a reader needs them, we will have abandoned the enterprise of shared resources on which the political economy of traditional libraries is built. I am, frankly, a pessimist about our institutional ability to develop allocation rules that are not founded in some sense of community and sharing. The only way I see to avoid these difficulties is for the cost of information to approximate zero.

"Why should most university presses in North America be so preoccupied with the scholarly monograph?"

Returning to the world I can predict, I must conclude that just-in-time models for carrying scholarly communication will not soon be decisively important in creating a viable market for infrequently read scholarly monographs. That conclusion prompts two sets of questions about the role of university presses, the traditional publishers of such monographs. The first set is about the policy view we take of infrequently read scholarly monographs.

  • Is it a good use of the talents of university press staff and of the financial resources they command to publish large numbers of scholarly monographs that do not have many readers? Is it not terribly wasteful to lavish these resources on books that command so little attention?
  • If the most compelling need for the infrequently read scholarly monograph relates to credentialing, then should we not look to other means for that purpose?[22] We know how to credential Ph.D. students as assistant professors without insisting that the infrequently read monographs they write — their dissertations — take the same published form as frequently read monographs. Can we not deal with credentialing associate and full professors much more sensibly than through the expensive fetish of the university press book that has too few readers to make it economically viable?

The second is about the economic place of such monographs in the larger enterprise of scholarly communication.

  • Even if we liberate university presses from a wasteful credentialing business, there will still be some important scholarly monographs that command too few readers to be viable in the marketplace. If the university subsidy now delivered to these books via the library is in jeopardy primarily because of very steep increases in the price of scholarly journals, why would one believe that an accounting shift within the university — away from libraries and toward university presses, as proposed by Sandy Thatcher — could actually happen in a way that would save the infrequently read scholarly monograph? Such an accounting shift neither addresses nor solves the problem. So I must continue to ask,[23] why should most university presses in North America be so preoccupied with the scholarly monograph, which at best represents but half of the scholarly communication enterprise, and be so little engaged with the scholarly journal?
  • Why do we create publishing enterprises and convene conferences that attempt to deal piecemeal with problems in scholarly communication we know to be inextricably linked?
  • Why should university press officers, or librarians, or university administrators, or faculty believe we can solve the problems of the scholarly monograph without simultaneously solving the problems of the scholarly journal?

I hope these questions will be addressed by the scholarly publishing community.

Scott Bennett is the Yale University Librarian. He received his A.B. degree from Oberlin College in 1960 and his Ph.D. from Indiana University in 1967. He completed his M.S. in library science at the University of Illinois in 1976. Dr. Bennett has published numerous papers on publishing and intellectual property, publishing history, textual editing, bibliography, preservation, and other topics. You may contact him by e-mail at


1. Throughout this paper, the term "infrequently read scholarly monograph" means a scholarly monograph that is used between five and fifty times over a 50-year period — i.e., between 0.1 and 1 times a year. General guidance for library cost studies is available in Gary M. Pitkin, Cost-Effective Technical Services (New York: 1989), which includes a substantial bibliography.return to text

2. Sanford G. Thatcher, "The Crisis in Scholarly Communication," The Chronicle of Higher Education, 3 March 1995, B1.return to text

3. There is, of course, a difference between operating costs and selling prices, but we should probably ignore the difference if we are to depend on commercial enterprises for access to information. The relationship of costs and prices in commercial enterprise is regulated by competition, unlike not-for-profit enterprises, where organizational mission is much more powerful in regulating the relationship between costs and prices.return to text

4. To do this, UMI has built a collection of film documents and preserves that collection over decades, much as a library does. These similarities should not obscure the basic fact that UMI's dissertation access business is founded on making copies, in direct response to requests for such copies, of something it is free to copy. Libraries are fundamentally in the business of shared access. They rarely make copies for readers, and their ability to do so is closely limited by copyright law. Libraries approximate UMI's mode of business only in photocopying articles for interlibrary loan use. Such activities generally account for less than 2 percent of total library circulation activity.return to text

5. The advantage of being able to support multiple uses simultaneously (limited only by manufacturing capacity) has relatively little value, of course, for infrequently read scholarly monographs.return to text

6. The unit cost basis for all factors included in the cost model, both for CRL and the Yale Library, is provided in Attachment 6. CRL's cataloging cost figure assumes original cataloging for every foreign dissertation. Because others at this conference have described the costs publishers incur in producing scholarly monographs, I have ignored the purchase price paid by the library for the book. I also excluded the cost of selecting the material in this model. The conditions of selection are quite different in my three cases. For UMI, they presumably are largely borne by thesis writers and graduate school administrators; for libraries they are a significant staff cost; and for digital delivery they are borne by the reader.return to text

7. The model treats capital costs (for the building and its shelving) as annual operating costs by amortizing them over thirty years (CRL) and thirty-five years (Yale).return to text

8. Except that the cost model for Yale's off-campus shelving allows for 24 hour (or less) delivery service from the shelving facility to one of Yale's library circulation desks. The circulation of both CRL and Yale materials includes a set of hidden costs in the work that readers perform in going to the library to get the material they want.return to text

9. See particularly the section on Managing Costs and Finances (27-35) and Appendix 2 of Donald Waters and John Garret, Preserving Digital Information. Report of the Task Force on Archiving of Digital Information, commissioned by the Commission on Preservation and Access and the Research Libraries Group, Inc. (Washington, D.C.: 1996). Project Open Book work also provides the basis for Waters's earlier essay, "Transforming Libraries through Digital Preservation," in Nancy E. Elkington, ed., Digital Imaging Technology for Preservation: Proceedings from an RLG Symposium held March 17 & 18, 1994 (Mountain View, Calif.: 1994), 115-27.return to text

10. The Project Open Book cost model focuses only on material used in a given year and does not attempt to model the cost of retaining that material in the local digital collection over an extended period of time.return to text

11. The terms "digital archive" and "depository library" were not associated with just-in-time and just-in-case provision of information in Waters and Garrett. But the same cost model is used in Waters's "Transforming Libraries" essay to explore briefly the utility of thinking about just-in-time and just-in-case services.return to text

12. This assumption is explicitly acknowledged to be conservative in Waters and Garrett, note 21, 47-48.return to text

13. See Waters and Garrett, note 22, 48, for a discussion of this.return to text

14. Waters recognized this would be the case, and therefore used his model to explore fruitfully a fundamentally different assumption about the extent of the collection that would be stored digitally: "if over the next decade storage in the digital archives is managed the same way as storage in conventional paper-based libraries — that is, if the number of volumes stored is the same as the number of volumes acquired — then the overall cost advantage would still favor the depository library" (Preserving Digital Information, 33).return to text

15. This advantage of the local collection exists only at comparable levels of use. One would assume that an individual title, held both at the Center and at Yale, would get more use at the Center because of its larger clientele. Assuming for instance that a title used five times in fifty years at Yale would be used twenty-five times at the Center, Yale's lowest per use cost would be $27.45 compared to the Center's cost of $22.51return to text

16. Of course the library bears additional costs for providing readers instruction in the use of licensed material, for administering licenses, and (with other units of the university) in providing a communications network and computer equipment for the use of licensed material.return to text

17. Indeed, the high per use cost reported for UMI in Attachment 5 reflects the fact that one cannot unbundle UMI's storage and access costs from the other, additional costs that are actually reflected in the reported prices.return to text

18. See Charles Knight, Passages of a Working Life (London: 1864), II, 66-67.return to text

19. Knight's millennial dream was founded, like ours, on favorable technology cost trends. He looked to mass-produced paper and rapid, steam-driven printing presses to transform the cost of providing information to readers, just as we look to the computer. See for instance Knight's landmark account of the fundamental technological transformations of his day in the monthly supplements to The Penny Magazine, September-December 1833. Knight published the Penny Magazine for the Society for the Diffusion of Useful Knowledge.return to text

20. Our experience to date is that licenses for digital bundles of information and access cost about a third more than their analog counterparts, in spite of highly favorable cost trends in technology. This being so, it is reasonable to speculate about a future in which the costs for digital information remain significant.return to text

21. One might ask why we focus our concern on the infrequently read scholarly monograph, when the much more frequently read mid-list book is also gravely threatened (see, for instance, The New York Times story "Middling (and Unloved) in Publishing Land," 18 August 1997, D1 & D6). And what leverage on the more general problem of creating an informed citizenry do we get by focusing on the protection of the infrequently read scholarly monograph — as opposed, say, to the protection of the National Endowment for the Humanities, the need to keep museum admission charges low, or the provision of high-quality public education? With threats to well-informed living so widespread, one needs to think carefully about a policy that gives a privileged (i.e., a subsidized) place to infrequently read scholarly monographs.return to text

22. For a more general discussion of credentialing, see Scott Bennett, "Re-engineering Scholarly Communication: Thoughts Addressed to Authors," Scholarly Publishing, 27 (1996): 185-96.return to text

23. For an expanded formulation of this question, see Scott Bennett, "Repositioning University Presses in Scholarly Communication," Journal of Scholarly Publishing, 25 (1994): 243-48.return to text

Attachment 1

UMI, Inc. Price Estimate
Softcover paper copy$36.00
Rush delivery25.00
Manufacturing cost
Price difference between microfilm/fiche and softcover paper-3.50
Estimated cost of film duplication/disserttion-10.00

Source: UMI, Inc. Order Form D (effective 1 January 1995)

Attachment 2

CRL Costs
Series 1Series 2Series 3
Cost elementUnit cost50 year cost75 year cost100 year cost
Acquisition (excluding purchase price)$9.06$9.06$9.06$9.06
Cataloging and shelf preparation$48.19$48.19$48.19$48.19
Circulation (assuming 5 transactions in 50 years)$19.78$98.90$148.35$197.80
Processing space$0.00$0.00$0.00$0.00
Operating costs (HVAC, maintenance, etc.)$0.00$0.00$0.00$0.00
SCENARIO 1: Cost per circulation (5 in 50 years)$33.43$29.61$27.71

Similar iterations of the model have been created for 10, 15, 20, 25, 30, 35, 40, 45, and 50 circulation transactions in 50 years. The cost per circulation for each of these 10 scenarios is summarized below.

SCENARIO 1: Cost per circulation (5 in 50 years)5$33.43$29.61$27.71
SCENARIO 2: Cost per circulation (10 in 50 years)10$26.61$24.70$23.74
SCENARIO 3: Cost per circulation (15 in 50 years)15$24.33$23.06$22.42
SCENARIO 4: Cost per circulation (20 in 50 years)20$23.19$22.24$21.76
SCENARIO 5: Cost per circulation (25 in 50 years)25$22.51$21.75$21.37
SCENARIO 6: Cost per circulation (30 in 50 years)30$22.06$21.42$21.10
SCENARIO 7: Cost per circulation (35 in 50 years)35$21.73$21.18$20.91
SCENARIO 8: Cost per circulation (40 in 50 years)40$21.49$21.01$20.77
SCENARIO 9: Cost per circulation (45 in 50 years)45$21.30$20.87$20.66
SCENARIO 10: Cost per circulation (50 in 50 years)50$21.15$20.76$20.57


  1. Operating costs include a 35% indirect cost allowance that covers items for which no estimate is made
  2. Circulation costs include long-distance delivery and OPAC maintenance costs

Attachment 3

Yale University Library Costs
Conventional shelvingSeries 1Series 2Series 3
Cost elementUnit cost50 year cost75 year cost100 year cosT
Acquisition (excluding purchase price)$12.00$12.00$12.00$12.00
Cataloging and shelf preparation$35.00$35.00$35.00$35.00
Circulation (assuming 5 transactions in 50 years)$1.75$8.75$13.13$17.50
Shelving (ADA Compliant)$1.72 $86.00$129.00$172.00
Processing space$0.06$3.00$4.50 $6.00
Operating costs (HVAC, maintenance, etc.) $1.16$58.00$87.00$116.00
Preservation $0.25 $12.50$18.75$25.00
SCENARIO 1: Total Cost$217.75$303.13$388.50
SCENARIO 1: Cost per circulation (5 in 50 years) $43.55$40.42$38.85

Similar iterations of the model have been created for 10, 15, 20, 25, 30, 35, 40, 45, and 50 circulation transactions in 50 years. The cost per circulation for each of these 10 scenarios is summarized below.

Series 1Series 2Series 3
SCENARIO 1: Cost per circulation (5 in 50 years)$43.55$40.42$38.85
SCENARIO 2: Cost per circulation (10 in 50 years)$22.65$21.08$20.30
SCENARIO 3: Cost per circulation (15 in 50 years)$15.68$14.64$14.12
SCENARIO 4: Cost per circulation (20 in 50 years)$12.20$11.42$11.03
SCENARIO 5: Cost per circulation (25 in 50 years)$10.11$9.48$9.17
SCENARIO 6: Cost per circulation (30 in 50 years)$8.72$8.19$7.93
SCENARIO 7: Cost per circulation (35 in 50 years)$7.72$7.27$7.05
SCENARIO 8: Cost per circulation (40 in 50 years)$6.98$6.58$6.39
SCENARIO 9: Cost per circulation (45 in 50 years)$6.39$6.05$5.87
SCENARIO 10: Cost per circulation (50 in 50 years)$5.93$5.62$5.46


Depository shelvingSeries 4Series 5Series 6
Cost elementUnit cost50 year cost75 year cost100 year cost
Acquisition (excluding purchase price)$12.00$12.00$12.00$12.00
Cataloging and shelf preparation$35.00$35.00$35.00$35.00
Circulation (assuming 5 transactions in 50 years)$1.75$8.75$13.13$17.50
Shelving (depository)$0.14$7.00$10.50$14.00
Processing space$0.06$3.00$4.50$6.00
Operating costs (HVAC, maintenance, etc.)$1.13$56.50$84.75$113.00
SCENARIO 1: Total cost$137.25$182.38 $227.50
SCENARIO 1: Cost per circulation (5 in 50 years)$27.45$24.32$22.75

Similar iterations of the model have been created for 10, 15, 20, 25, 30, 35, 40, 45, and 50 circulation transactions in 50 years. The cost per circulation for each of these 10 scenarios is summarized below.

SCENARIO 1: Cost per circulation (5 in 50 years)$27.45$24.32$22.75
SCENARIO 2: Cost per circulation (10 in 50 years)$14.60$13.03$12.25
SCENARIO 3: Cost per circulation (15 in 50 years)$10.32$9.27$8.75
SCENARIO 4: Cost per circulation (20 in 50 years)$8.18$7.39$7.00
SCENARIO 5: Cost per circulation (25 in 50 years)$6.89$6.26$5.95
SCENARIO 6: Cost per circulation (30 in 50 years)$6.03$5.51$5.25
SCENARIO 7: Cost per circulation (35 in 50 years)$5.42$4.97$4.75
SCENARIO 8: Cost per circulation (40 in 50 years)$4.96$4.57$4.38
SCENARIO 9: Cost per circulation (45 in 50 years)$4.61$4.26$4.08
SCENARIO 10: Cost per circulation (50 in 50 years)$4.32$4.01$3.85

Attachment 4

Project Open Book Cost Information

Table 1. Costs of the Digital Archives
Cost FactorsCosts per Volume in Year 1
Device Costs$0.97
Device Maintenance0.40
Operations Costs0.40
Media Costs0.25
Media Refreshment and Data Migration0.49
Storage Systems Engineering0.04
Storage Management Overhead0.03
Total Storage Costs per Volume$2.58
Document Server$2.13
Server Maintenance0.88
Server Operations0.88
Access Software1.22
Software Maintenance 0.50
Access Systems Engineering0.04
Access Management Overhead0.03
Total Access Costs per Volume$6.72
Table 2. Depository of Library Costs
Cost FactorsCosts per Volume in Year 1
Total Storage Costs per Volume$0.21
Total Access Costs per Volume$4.40
Table 3. Projected Costs Per Volume Over 10 Years
Year 1Year 4Year 7Year 10
Depository Library
Depository Storage Costs Per Volume$0.21$0.24$0.27$0.30
Depository Access Costs Per Volume$4.40$4.95$5.57$6.27
Digital Archives
Digital Storage Costs Per Volume$2.58$1.73$1.18$0.82
Digital Access Costs Per Volume$6.72$4.84$3.60$2.79
Table 4. Costs for All Volumes Stored and Used
Year 1Year 4Year 7Year 10
New Volumes200,000200,000200,000200,000
Depository Library (volumes stored=new volumes)
Estimated annual use (15% of volumes)30,00030,00030,00030,000
Depository Storage Costs for New Volumes$42, 769$48,110$54,117$60,874
Depository Access Costs for Volumes Used$132,090$148,583$167,136$188,005
Total Depository Storage and Access Costs$174,859$196,693$221,253$248,879
Digital Archives (volumes stored=volumes used)
Estimated annual use (20% of volumes)40,00040,00040,00040,000
Digital Storage Costs for Volumes Used$103,208$69,371$47,207$32,763
Digital Access Costs for Volumes Used$268,870$193,400$143,977$111,785
Total Digital Storage and Access Costs$372,078$272,771$191,184$144,549
Difference: Depository - Digital($197, 219)($66, 078)$30,069$104, 331

Copied as a matter of Fair Use from Donald Waters & John Garrett, Preserving Digital Information. Report of the Task Force on Archiving of Digital Information Commissioned by the Commision on Preservation and Access & The Research Libraries Group, Inc. (Washington D.C. 1996)

Attachment 5

Cost Comparisons

Assuming 1 Circulation within a 10 Year Period

1 circ. in 10 years, in:Year 1Year 2*Year 3Year 4Year 5Year 6Year 7Year 8Year 9Year 10
Project Open Book
Depository Library (Series 1)
SERIES 1 TOTAL$6.95 $7.13$7.31$7.50$7.70$7.91$8.12$8.34$8.58$8.82
Digital Archives (Series 2)
SERIES 2 TOTAL$22.09$21.37$20.75$20.21$19.73$19.32$18.97$18.66$18.40$18.16
UMI, Inc. (Series 3)
SERIES 3 TOTAL$45.60$47.50$49.40$51.38$53.43$55.57$57.79$60.10$62.51$65.01
Center for Research Libraries (Series 4)
SERIES 4 TOTAL$19.20$20.00$20.80$21.63$22.50$23.40$24.33$25.31$26.32$27.37
Yale Library, depository shelving (Series 5)
HVAC, etc.$1.08$1.13$1.18$1.22$1.27$1.32$1.37$1.43$1.49$1.55
SERIES 5 TOTAL$3.14$3.27$3.40$3.54$3.68$3.83$3.98$4.14$4.30$4.48

* Base year for calculating 4% inflation on UMI, CRL, and Yale Library costs.

Attachment 6

Unit cost model for CRL and Yale Library


CRLYale Library
One-time cost per volume
Acquisition activities$9.06$12.00Allowance
Cataloging and shelf preparation$48.43$35.00Allowance
Annual cost per volume
Circulation (out and in)$19.54$1.75OCS Study, Doc. 9; includes OPAC costs
Shelving space (CRL)$0.22
Shelving space (ADA compliant)$1.72Amortized construction cost/vols. per sq. foot
Shelving space (depository)$0.14Amortized construction cost/vols. per sq. foot
Amortized proc. & admin space/yr.IDC$0.06Amortized cap. cost per net sq. foot*space per vol.
Operating costs (ADA shelving)IDC$1.12HVAC, etc.*ratio/vols. per sq. foot
Operating costs (depository shelving)IDC$0.01HVAC, etc.*ratio/vols. per sq. foot
Operating costs (processing & admin.)IDC$0.04HVAC, etc.*ratio*space per vol.
Cost elements for space
Capital costs
Construction cost/net sq.foot$70$300Allowance
Amortization period30 years35 years
Interest rate6%8%
Amortized cap. cost/net sq.foot$20.69$25.74Amortization table
Vols./sq.foot (CRL)100
Vols./sq.foot (ADA compliant)15OCS Study, Doc. 1
Vols./sq.foot (depository)185OCS Study, Doc. 1
No. of volumes2,000,000
Processing & admin. spaceIDC5,000 nsfOCS design, in progress
Processing & admin. space/vol.IDC0.0025Total space/number of volumes
Operating costs
HVAC, maintenance, etc./gsfIDC$12Allowance
Ratio gsf/nsf1.0981.4Allowance


CRL cost information provided by Donald Simpson, CRL President, in August 1997 correspondence. CRL accounts for many cost factors that are itemized by Yale by applying a 35% indirect cost (IDC) rate to its core activities. Yale Library cost information based on estimates, allowances, and formulas provided above.